opus - Jul 2018 - OPUS on cortex M4 (Nicolas Ehrenberg)

Thanks for the reply. 
For my application I unfortunately need a better signal reconstruction. It's
not necessarily a problem that the constant DC voltage is removed, but the audio
signal will need to be more exact because it's also studied visually. To be
more exact, I need to record and transmit audio data recorded from animals
(mostly birds).

Are there ways to achieve a more similar output signal compared to the input
signal?

As I already mentioned I tried different complexities which did not lead to much
improvement. If I input other data values,  e.g, 2000 +- 500, or 2000 with
occasional dips and peaks, I still do not get a reasonably good visual signal
reconstruction.

Also, I'm not quite sure if my opus implementation works as it should yet. I
can run the encoder with OPUS_APPLICATION_RESTRICTED_LOWDELAY, but If I try to
run it with OPUS_APPLICATION_VOIP or OPUS_APPLICATION_AUDIO the opus_encode
function returns just 2 (size of encoded samples) and the opus decode just
returns 0's. I would imagine to get the best audio reconstruction with
OPUS_APPLICATION_AUDIO.
What could be the reason for OPUS_APPLICATION_AUDIO not working?

Best,
Nick


________________________________________
From: opus [opus-bounces at xiph.org] on behalf of opus-request at xiph.org
[opus-request at xiph.org]
Sent: Saturday, June 30, 2018 2:00 PM
To: opus at xiph.org
Subject: opus Digest, Vol 113, Issue 8

Send opus mailing list submissions to
        opus at xiph.org

To subscribe or unsubscribe via the World Wide Web, visit
        http://lists.xiph.org/mailman/listinfo/opus
or, via email, send a message with subject or body 'help' to
        opus-request at xiph.org

You can reach the person managing the list at
        opus-owner at xiph.org

When replying, please edit your Subject line so it is more specific
than "Re: Contents of opus digest..."


Today's Topics:

   1. Re: OPUS on cortex M4 (Timothy B. Terriberry)


----------------------------------------------------------------------

Message: 1
Date: Fri, 29 Jun 2018 14:41:07 -0700
From: "Timothy B. Terriberry" <tterribe at xiph.org>
To: Ehrenberg Nicolas <enicolas at student.ethz.ch>, "opus at
xiph.org"
        <opus at xiph.org>
Subject: Re: [opus] OPUS on cortex M4
Message-ID: <4ad369bd-a04a-de38-67fe-855565891973 at xiph.org>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed

Ehrenberg  Nicolas wrote:
>      for(uint16_t i = 0; i < test_data_size; i++)
>      {
>      test_data[i] = 100;
>      }
>
> [snip]
>
> The decoded values start out at 0, go to -16 up, then up to 135 and then
> slowly down to the 17 (see attached file).
> Furthermore, I do not see much of a change if I increase the complexity.
You're encoding a signal with a constant (DC) offset. opus_encode()
includes a 3 Hz highpass filter to remove the (inaudible) DC component,
because it screws up other parts of the codec (mainly it would create
MDCT leakage that is completely unmasked). So you shouldn't expect to
get your constant values of 100 back.

In general the numeric values of the output aren't guaranteed to
resemble the numeric values of the input. The codec will just try to
make the two signals sound the same to human ears.


------------------------------

Subject: Digest Footer

_______________________________________________
opus mailing list
opus at xiph.org
http://lists.xiph.org/mailman/listinfo/opus


------------------------------

End of opus Digest, Vol 113, Issue 8
************************************

Ehrenberg  Nicolas wrote:
> Also, I'm not quite sure if my opus implementation works as it should
yet. I can run the encoder with OPUS_APPLICATION_RESTRICTED_LOWDELAY, but If I
try to run it with OPUS_APPLICATION_VOIP or OPUS_APPLICATION_AUDIO the
opus_encode function returns just 2 (size of encoded samples) and the opus
decode just returns 0's. I would imagine to get the best audio
reconstruction with
In either of these modes, the encoder adds enough delay to allow 
switching between the CELT and SILK layers. That may be more than a full 
frame for small frame sizes (5 ms or less). The result is that the first 
frame out of the encoder will be just the padding (usually pure 
silence), which is consistent with a 2 byte frame and all 0's from the 
decoder.

Also, even with OPUS_APPLICATION_RESTRICTED_LOWDELAY, there is still 
some delay added. See the opus_encoder_ctl OPUS_GET_LOOKAHEAD() to query 
the delay actually in use.

After that, pretty much all I can suggest to improve the "visual" 
appearance of the signal is to increase the bitrate. But even if you go 
all of the way up to 255 kb/s per channel, it's still a lossy codec 
designed for human ears, not spectogram plots.

2000 +- 500 still has a large DC component that will get removed.